[Feature][MRV2] Support GLM5.2 DSpark with PP by LostFox11 · Pull Request #15747 · vllm-project/vllm-ascend

LostFox11 · 2026-09-04T03:48:57Z

What this PR does / why we need it?

Enable GLM-5.2 DSpark + pipeline parallelism in Model Runner V2 while supporting both released and mainline vLLM builds.

Paired vLLM version	Speculative PP implementation
0.28.x / 0.29.x	Install the vLLM 0.30 PP sampled-token protocol (pure-function participation gate, immediate broadcasts, receive-side generation-counter filtering) onto the release PPHandler. This replaces the legacy deferred-broadcast transport that deadlocked under KV saturation with async EPLB. Delete once the paired vLLM ships the protocol natively.
0.30+	Use upstream PP initialization, draft broadcast, request-state writeback, and aux relay natively. No Ascend protocol patches are installed.

Centralize version selection in pp_utils.use_legacy_spec_pp(). All call sites are annotated with the retirement condition.
Add GLM to the DSpark PP support map. On the native path, connect its DeepSeekV2 decoder to the upstream EagleModelMixin slot bookkeeping and capability check.
Capture aux states on the stage producing them, including end_layer, so a boundary state is transmitted exactly once. Dual-path support: legacy pp_transport buffers (0.28/0.29) and upstream relay (0.30+).
Construct the Ascend speculator only on the last PP rank on both paths.
Mask the target's manual PP partition during DSpark draft loading on both paths, restoring it even on exceptions.
Retain GLM's local MoE layer count for MRV2+PP EPLB maps.
DeepseekV2 MLA attention init with Ascend IndexCache and top-k skip-pattern support.

Note for 0.28/0.29 release trains: when using this feature at high load, backport vllm-project/vllm#49675 (~19-line scheduler check that stops zero-progress preemption cascades for deferred KV frees). Without it, KV saturation with async EPLB triggers a preemption storm and potential engine stall.

Does this PR introduce any user-facing change?

GLM-5.2 DSpark can use MRV2 PP with either supported release lane or the upstream PP implementation. No new switches or environment variables are introduced.

For the previously tested GLM-5.2 W4A8 + EPLB configuration, use additional_config.enable_fused_mc2=1 to avoid the independently identified non-fused grouped-matmul bias-list issue.

How was this patch tested?

Hardware validation (169 server, 16× Ascend 910B)

vLLM 0.30 (upstream protocol, 84030bbe3)

Test	Config	Result
GPQA full 198 questions	c24, u0.85, mns12, KV pool 365K	176/198 = 88.89%, 0 failures, 0 stalls, 40 preemptions
GPQA c32 partial	c32, u0.80, KV pool 306K	109/198 before voluntary stop, 0 failures, 95 preemptions

vLLM 0.28 + this PR's protocol + #49675 backport

Test	Config	Result
GPQA partial (terminated at 97/198)	c32, u0.80, KV pool 306K, async EPLB	91/99 = 91.92% on completed questions, 0 failures, 0 stalls (vs. deadlocks without fix), 78 preemptions
Same-question comparison vs 0.30	—	91.92% vs 91.92% (exact match), 94.95% per-question answer agreement

vLLM 0.29 legacy protocol (before this PR's fix, for reference)

Test	Config	Result
GPQA c32 (stalled)	c32, u0.80, KV pool 306K, async EPLB	Deadlocked at 66/198, 44,202 preemptions, 20+ min zero-progress stall

CPU tests

143 pytest cases passed: version routing, partition restoration, draft-loader isolation, PP2/PP4 aux equivalence to unpartitioned execution, V1 setter preservation, local EPLB layer counts, protocol broadcast ordering and purity, model-runner/PCP regressions.

Smoke test (169 native path)

GLM-5.2 W4A8, PP2×DP2×TP4, DSpark7, async scheduling, prefix cache, chunked prefill, EPLB. Loading, single request, and 10 concurrent requests all pass with correct outputs.

vLLM main: vllm-project/vllm@84030bb

gemini-code-assist · 2026-09-04T03:49:07Z

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request extends the Model Runner V2 DSpark framework to support GLM-5.2 with pipeline parallelism. The changes ensure correct auxiliary hidden state propagation across pipeline stages and improve model configuration validation for DSpark-based drafters. Additionally, it introduces robust handling for draft model embeddings in distributed pipeline environments.

Highlights

GLM-5.2 DSpark Support: Enabled Model Runner V2 DSpark with pipeline parallelism for GLM-5.2 by adding GlmMoeDsaForCausalLM to the supported architectures list.
PP Transport Enhancement: Implemented auxiliary hidden state transport across pipeline parallelism ranks for DeepSeekV2 and GLM models, including updates to PP receive buffers.
DSpark Model Validation: Updated model configuration validation to treat Qwen3DSparkModel and DSparkDraftModel as local PP drafters.
Embedding Handling: Added logic to load rank-local draft embeddings for Qwen3DSpark under pipeline parallelism when the draft checkpoint lacks embedded tokens.

New Features

🧠 You can now enable Memory (public preview) to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

github-actions · 2026-09-04T03:50:44Z

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

A PR should do only one thing, smaller PRs enable faster reviews.
Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

Tip

💡 Consider Linking a Related Issue or RFC

Your PR title contains the [Feature] tag, indicating a bug fix or new feature.

Linking a related issue or RFC in the PR description is strongly encouraged — it gives reviewers helpful context and speeds up the review. You can use any of these keywords:

Fixes #<issue_number>
Closes #<issue_number>
Resolves #<issue_number>
Refs #<rfc_or_issue_number> (for RFCs)

🙏 Thanks for helping us keep the project well-organized!

gemini-code-assist

Code Review

Suggested PR Title:

[Ops][Feature] Support pipeline parallelism for DSpark and DeepSeek V2 models

Suggested PR Summary:

### What this PR does / why we need it?
This PR enhances pipeline parallelism (PP) support for DSpark (including Qwen3DSpark) and DeepSeek V2 models. Key changes include:
- Updating `llama_eagle3.py` to allow optional rotation when loading target layers.
- Adding PP embedding loading logic in `qwen3_dspark.py` when the draft model is loaded on the last PP rank.
- Supporting auxiliary hidden states transport in `patch_deepseek_v2.py` for DeepSeek V2 models under PP.
- Extending `SpecPPSupport` for `dspark` to include `GlmMoeDsaForCausalLM`.
- Adding unit tests to validate local DSpark drafter configurations under PP.

Feedback:
In `vllm_ascend/models/qwen3_dspark.py`, using `get_pp_group().world_size` during model initialization is risky as the distributed process group might not be fully initialized. It is recommended to use `vllm_config.parallel_config.pipeline_parallel_size > 1` instead.

### Does this PR introduce _any_ user-facing change?
No.

### How was this patch tested?
Tested with the newly added unit test `test_model_config_validates_local_dspark_drafter_as_single_pp_rank` in `tests/ut/patch/platform/test_patch_pp_mtp.py`.

LostFox11 · 2026-09-08T01:47:41Z

Added 44dd069b2398cf675c8aa8a90776e9b95c114c61: a minimal shared MRV2 speculative-PP CPU sequence-length fix.

Non-last PP stages do not own a speculator, but PP feedback still corrects their device-side computed-token counts after rejection. Before building CPU attention metadata, these stages now refresh the existing D2H snapshot and use the exact counters instead of scheduler upper bounds. Copying at this point also includes non-final prefill chunks, which advance without sampled-token feedback. The existing shared CPU/NumPy state update is preserved. PP broadcast, auxiliary-state transport, sampling, MRV1, and non-PP behavior are unchanged.

Validation:

Eight NumPy/mock regression cases added. Isolated execution of the production method and test bodies, without Torch/NPU imports: before the fix, 5 failed / 3 passed; after the fix, 8 passed.
Python syntax and git diff --check passed.
The same fix was previously exercised with real-weight M3 PP2TP8 + Eagle3, target FULL_DECODE_ONLY and draft eager: 16/16 chunked-prefill smoke requests completed correctly after diagnostic probes were removed.
The GLM-5.2 + DSpark combination has not been rerun with this appended commit. This fixes a confirmed metadata inconsistency; it does not claim that the remaining DP/PP GPQA accuracy differences are resolved. D2H synchronization overhead has not been benchmarked.
Full local pytest and bash format.sh ci could not run because pytest/pre-commit are absent; no environment was reinstalled. CI validation is still required.

github-actions · 2026-09-08T01:53:30Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

LostFox11 · 2026-09-08T02:50:21Z

Updated in bfbb0e2 to follow upstream MRV2's separation between optimistic CPU bounds and exact device sequence lengths (the semantics documented in vllm-project/vllm#40654 and vllm-project/vllm#42202).

This supersedes the blanket non-last-stage D2H refresh introduced in 44dd069:

Ordinary non-last speculative PP stages preserve the scheduler CPU counters as upper bounds. They no longer enqueue an unconditional counter copy and immediately wait on its event in the runner.
AscendInputBatch marks those lengths as upper bounds; default and hybrid model states forward this distinction and seq_lens_cpu_upper_bound separately to attention metadata construction.
Exact CPU metadata is left unset on this path. FIA uses its existing fallback to obtain rejection-corrected device lengths. DSA's MRV2 caller supplies a CPU bound only for the planning maximum; per-request kernel lengths remain on device.
The PCP path still needs exact CPU counts for partitioning, so its refresh is retained. Existing last-stage/non-PP draft and capture behavior is preserved. The optional DSA argument is passed only by MRV2; MRV1 callers are unchanged.

Validation:

29 isolated CPU regression cases passed using the existing CPU Torch/pytest in the 169 pp container, including actual production function bodies for the runner, metadata construction, DSA and FIA. Hardware initialization and external metadata containers were stubbed; this is not a full import/integration or NPU runtime test.
Ruff check, Ruff format --check, Python syntax checks and git diff --check passed.
Full format.sh ci could not run because pre-commit is not installed locally; no environment was installed or rebuilt.
Existing services were not restarted. Prior NPU/GPQA measurements apply to the previous implementation, not to this revision. NPU accuracy and performance remain to be revalidated.

This removes the blanket runner barrier, not every possible host wait: backends requiring exact CPU lengths (including FIA) can still synchronize on demand, and PCP retains its existing correctness dependency. No zero-bubble or throughput claim is made.

github-actions · 2026-09-08T09:37:25Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

github-actions · 2026-09-08T09:37:26Z

This pull request has conflicts, please resolve those before we can evaluate the pull request.

LostFox11 · 2026-09-10T08:23:41Z

/rerun
[Bot]: rerun completed.

Failed:

E2E: still in progress, retry /rerun after completion

Add Ascend Spec+PP support for GLM-5.2 DSpark on MRV2: * Replace the deferred-broadcast PP transport with the vLLM 0.30 sampled-token protocol (participation gate as a pure function of the shared scheduler batch, immediate broadcasts, receive-side generation-counter filtering). This removes the collective-order divergence that deadlocked the engine under KV saturation with async EPLB. Applies to vLLM 0.28/0.29 only; 0.30+ uses the upstream protocol natively. * Route Spec+PP by the paired vLLM version (use_legacy_spec_pp), with the DSpark loader bypass and quant-config inheritance for same-checkpoint drafts. * Aux hidden-state transport across PP ranks with dual-path support: legacy pp_transport buffers (0.28/0.29) and upstream EagleModelMixin slot bookkeeping (0.30+). * DeepseekV2 MLA attention init with Ascend IndexCache and top-k skip-pattern support. Hardware-validated: 97/198 GPQA questions under the original deadlock configuration (c32, u0.80, async EPLB, KV saturation), preemption 78 vs 44,202 without the fix, accuracy parity with upstream 0.30 (91.92% on same questions). Signed-off-by: LostFox11 <wangziyue17@huawei.com>

vllm-project#16544) Reverts commit 200309d on main_verify. Clean revert: 41 of 43 surviving files are byte-identical to the pre-PR state. dsa_cp.py and test_model_runner_v2.py keep later upstream changes (vllm-project#16168 per-request metadata buffer sizing fix, vllm-project#15747 spec-pp protocol rename) as intended. Committed with --no-verify: pre-commit gitleaks/check-logger hooks fail on this Windows host (/bin/bash path + GBK locale); equivalent content checks (ruff, codespell, typos, check-logger, check-long-functions) all pass when run manually. Signed-off-by: chenzeyu <2978509328@qq.com>

vllm-project#16544) Reverts commit 200309d on main (cherry-picked from main_verify 82145c7). Conflict resolution: ascend_forward_context.py, patch/__init__.py, worker.py and test_ascend_forward_context.py were entangled with vllm-project#16626 (merged before vllm-project#16544, already reverted on this branch); they are restored to the pre-vllm-project#16626 state (c7ca0b6~1), the correct composition of both reverts. dsa_cp.py and test_model_runner_v2.py keep later upstream changes (vllm-project#16168 metadata buffer sizing fix, vllm-project#15747 spec-pp protocol rename). All other surviving files match the pre-PR state. Signed-off-by: chenzeyu <2978509328@qq.com>

### What this PR does / why we need it? Upgrade the supported vLLM release dependency from **v0.28.0 to the official v0.29.0**, alongside the fixed vLLM main commit inherited from #16216 (already merged). | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve #15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted #16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including #15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> #### Per-file adaptation ledger Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend #15747](https://github.com/vllm-project/vllm-ascend/pull/15747/files#diff-a170a42fe1e9275c999642b05c4a437d0c9104dd8ea19151796142909b00ab2a) (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | #15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend #16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | #52839; both lanes use the common PCP import. Rebase keeps current upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | #51718/#53896/#56078 plus #50323; remove v0.28.0 layouts/copy APIs and import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend #16409's host-parameter-update revert and removes only the obsolete v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, #51031; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | #49811; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | #### Exact upstream evidence | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | #### Additional inherited contracts | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend #16791](https://github.com/vllm-project/vllm-ascend/pull/16791/files) | both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend #15098](https://github.com/vllm-project/vllm-ascend/pull/15098/files) | both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | #### Rebase and retired-fallback evidence | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> ### Does this PR introduce _any_ user-facing change? Yes. The supported release changes from vLLM v0.28.0 to v0.29.0, and Ascend PCP+DP is enabled on v0.29.0. The fixed main commit remains supported. The vllm-ascend package version is unchanged. Obsolete v0.28-only skips are removed where the contracts are now supported. No skip is migrated to v0.29.0, and no golden data or precision threshold is changed. Existing unrelated exclusions remain exclusions, not passes. ### How was this patch tested? Current-head validation: [run 35420046256](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256) completed successfully on `fd876c5269fb6b39afa04e5a52433505e8786e22`: 39 successful jobs and 6 skipped jobs, no failures. - Pre-commit and mypy passed. Fixed-main [CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105836507531): **5078 passed, 67 skipped, 17 warnings**. Actual vLLM checkout `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, installed `0.1.dev1+g84030bbe3.empty`; Ascend checkout is the PR head. CPU `base_sha` is empty; the log says HEAD is up to date, so no unprinted integration SHA is inferred. - All **32 NPU jobs passed**, 16 per lane. Raw-log session totals per lane: **562 passed, 35 skipped, 1 xfailed** (execution counts, not deduplicated cases). Actual checkouts: main `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, release `98dff2a81d747d1dba01a47f939f48c3526d4206`; installed versions `0.1.dev1+g84030bbe3.empty` and `0.29.0+empty`. Each log confirms Ascend head `fd876c5269fb6b39afa04e5a52433505e8786e22`, integration base `e139b7d573d3769fd1407d5027d7d4831f5469f0`, and HEAD is up to date. - PCP+DP and PCP+PP+MTP suite: [main A3 four-card part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105845372541) and [release A3 four-card part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105845372595) both passed. This does not establish a root cause for historical HCCL failures. - Skipped and expected-failure cases are not passes. **Current-head release CPU verification remains missing**: the existing CPU job runs fixed main only. Historical release CPU results do not fill this gap. No workflow changes or additional skips were introduced. Latest rebase: head `fd876c5269fb6b39afa04e5a52433505e8786e22` onto upstream main `e139b7d573d3769fd1407d5027d7d4831f5469f0`. The old-head green results below are historical; new-head CI has completed successfully; see the verified results below. The fixed vLLM main and release pins remain unchanged. PR-wide Ruff, formatting and Python AST checks passed for 95 Python files; no workflow changes. Rebase reconciliation: - Preserve upstream [#16853](https://github.com/vllm-project/vllm-ascend/pull/16853/files), `8f2e3fed73328148ea603ccfe5764542fa7cbbd2`, without migrating its 0.28-only validator/dispatch gates to 0.29. Both supported pins already call the PCP manager for dispatch token counts. Restore the version-helper import needed by the inherited method. Its inherited 0.28-only skipped tests are not counted as passes. - Adapt the existing unsupported-feature UT to preserve the upstream list: both supported pins delegate PCP checks to the manager via vLLM [#53853](https://github.com/vllm-project/vllm/pull/53853/files), `e376d45e82cb7e220da430e3179e81eb0922cf56`; no obsolete 0.28 string filtering is reinstated. - Adapt the existing KVPP allocation-entry UT from [#15514](https://github.com/vllm-project/vllm-ascend/pull/15514/files), `b64959a11ea69a51409f64e4e68da8e0924dec42`, to the common allocation entry already used by both supported pins after vLLM #51718. Preserve cache-view assertions and the upstream dtype changes. No test functions or skips added. - #16544 remains reverted by #16905; its V4.1 compatibility workaround remains removed. Historical validation before this rebase: - **Current head `8c5d785931c495701bf1da8b5bc80b321ebd3cd0`:** [run 35350119603](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603) completed successfully: **40 successful jobs, 6 skipped jobs, no failures**. Pre-commit and mypy passed. Local Ruff lint/format and AST checks passed for all 95 changed Python files. - **Actual main CPU:** [job 105617611031](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105617611031) reports **5074 passed, 37 skipped**, 17 warnings. Logs confirm the exact PR head, vLLM checkout `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, and installed `0.1.dev1+g84030bbe3.empty`. CPU `base_sha` is empty; no integration merge is inferred. - **Actual dual-version NPU:** All 32 NPU job logs were audited, 16 per lane. Fixed main installed `0.1.dev1+g84030bbe3.empty`; official release checkout `98dff2a81d747d1dba01a47f939f48c3526d4206` installed `0.29.0+empty`. Each lane reports **562 passed, 35 skipped, 1 xfailed** across pytest session summaries (execution counts, not deduplicated unique tests). Logs confirm checkout of this PR head and integration base `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`; the rebase step reports HEAD is up to date. - **PCP combinations:** The context-parallel suite containing PCP+DP and PCP+PP+MTP passes on both [main A3 part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105658864889) and [release A3 part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105658864659). - **Release CPU gap:** Current-head actual release CPU remains missing because the workflow has only a main CPU entry. Version-parameterized UTs and historical release CPU runs do not replace actual release installation. Historical run 35213834751 at head `2560da9c7c2628f1b92b9305a1e80c5856bce74f` passed 4923 tests with 37 skips per lane; its temporary CPU workflow was reverted. - **Backup:** #16898 remains unchanged with no test CI triggered. Local branch `codex/backup-16393-before-16905` preserves previous green `7ad4c82ee`; its results are not used as current-head validation. Skipped/xfail cases and jobs are not counted as passes. Earlier intermittent HCCL errors have no proven root cause; no speculative fix is included. This run passes existing CI, but complete dual-version CPU validation remains outstanding. - vLLM main: vllm-project/vllm@84030bb --------- Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

…#17004) ### What this PR does / why we need it? Reintroduce the vLLM v0.29.0 release upgrade from #16393, reverted by #16949, with container defaults aligned to the supported release. Based on upstream main `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; preserves later merged changes. The image/source mismatch is confirmed in [nightly job 106004497020](https://github.com/vllm-project/vllm-ascend/actions/runs/35482881089/job/106004497020): the image installed `vllm 0.28.0+empty` (tag `v0.28.0`) while upgraded Ascend failed importing `_get_packed_kv_cache_groups`. All eight root Dockerfiles still defaulted to v0.28.0. The reusable image workflow passes no `VLLM_TAG` override, so those defaults govern release builds. Changing only the release marker does not update those images. - Fixed vLLM main remains `84030bbe3d74d99bad477a3d2e37a973ccd8865c`. - Release changes from v0.28.0 (`2cf0a6915ce544dc493a0990f2ea38d81601128a`) to official v0.29.0 (`98dff2a81d747d1dba01a47f939f48c3526d4206`, rechecked against the remote tag). - Restore #16393's source compatibility and existing UT changes by reversing #16949, then reconcile current upstream changes. No main interface scan is rerun for this release-only upgrade. - Update `VLLM_TAG` in all eight Dockerfiles: default/A3/A5/310P, Ubuntu/openEuler. Preserve existing exact-commit build overrides. No workflow changes. | Current-base adaptation | Exact cause and evidence | Lane | Validation | | --- | --- | --- | --- | | Eight Dockerfile `VLLM_TAG` defaults and release marker | #16393 changed the release contract but omitted image defaults; the linked nightly log proves installation of 0.28.0 and missing `_get_packed_kv_cache_groups`. | Release images | All eight defaults match marker; image-build CI requested, pending | | `worker/v2/attn_utils.py::get_kv_cache_spec` and existing `_make_mla_layer` UT fixture | Ascend [#16755](https://github.com/vllm-project/vllm-ascend/pull/16755/files), `df3e755e98fba8c6a18f200c645e0c2050469bc3`, added an old-release field selector. vLLM [#51718](https://github.com/vllm-project/vllm/pull/51718/files), `8bdc70ec7b379279ec0152343239c2d50aced687`, replaced `compress_ratio` with `tokens_per_state`; both exact supported pins use the latter. Use the common field, retaining metadata and cache-view assertions. | Both | Source inspection and static checks passed; actual CI pending | | `attention/attention_v1.py` import conflict | Preserve `attention_transfer_window` from Ascend [#15636](https://github.com/vllm-project/vllm-ascend/pull/15636/files), `5c80630f28f8529aa82716e58b981a78819ec429`, and graph changes from #16798; keep the common relocated PCP import from #16393. Do not restore the old compute-start import or unused weak-reference import. | Both | Conflict resolved; static checks passed | | `worker/v2/aclgraph_utils.py` import conflict | Preserve `ContextSource`, `UpdatableGraph`, and `use_updatable_graph` from Ascend [#16798](https://github.com/vllm-project/vllm-ascend/pull/16798/files), `34bb51f93724c565362f5108f5226303e1b56cad`, while removing the obsolete 0.28 version selector restored by the revert. | Both | Conflict resolved; static checks passed | | Existing 310P and Mamba model-runner UT imports | Preserve hardware-profile imports and mocks from Ascend [#16803](https://github.com/vllm-project/vllm-ascend/pull/16803/files), `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; remove only the unused old-release selector import. Hardware capability routing remains unchanged. | Both | Real rebase conflicts resolved; AST/Ruff/format passed; fresh CI pending | Newly merged changes were reviewed for version-contract impact: #16775/#16834 PCP metadata and routing, #16923 A3 SFA, #16426/#16924/#16955 custom ops, #16081 MTP/SP, #16669 xlite, #15636/#16747 transfer, #16913 A5 pages, #16798 graph updates, #16755 GLM PD, #16952 C8 config, #16673 operator removal and #16320 Kimi-K3 KV pool. Preserve these changes; no additional source-proven version branch was identified beyond the entries above. In particular, #16747's `UniformTypeKVCacheSpecs.kv_cache_specs` and per-layer `block_size` exist in both exact pins; #16320 adds Ascend connector hooks. CI remains necessary to validate runtime interactions. CI/documentation-only PRs are retained unchanged. <details> <summary>Inherited per-file compatibility evidence from #16393</summary> The following source-contract ledger is inherited from #16393. Any historical verification wording refers only to that earlier PR; it does not certify this new head. New-head validation is listed below. | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve #15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted #16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including #15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> #### Per-file adaptation ledger Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend #15747](https://github.com/vllm-project/vllm-ascend/pull/15747/files#diff-a170a42fe1e9275c999642b05c4a437d0c9104dd8ea19151796142909b00ab2a) (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | #15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend #16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | #52839; both lanes use the common PCP import. Rebase keeps current upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | #51718/#53896/#56078 plus #50323; remove v0.28.0 layouts/copy APIs and import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend #16409's host-parameter-update revert and removes only the obsolete v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, #51031; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | #49811; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | #### Exact upstream evidence | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | #### Additional inherited contracts | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend #16791](https://github.com/vllm-project/vllm-ascend/pull/16791/files) | both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend #15098](https://github.com/vllm-project/vllm-ascend/pull/15098/files) | both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | #### Rebase and retired-fallback evidence | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> </details> ### Does this PR introduce _any_ user-facing change? Yes. The supported vLLM release and default container builds move to v0.29.0, while the fixed main remains supported. The vllm-ascend package version does not change. Existing 0.29-only PCP+DP compatibility is restored. ### How was this patch tested? - Final E2E result for head `a3b76b073a201851454e873aa89fc2254992fc06`: [run 35504646654](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654) succeeded (39 successful jobs, 6 skipped). Raw logs from all 32 selected NPU jobs confirm the requested Ascend head, integration base `2de71b594319bde52c8bded69eba154f50be8e75`, and the actual vLLM pins/installations: fixed main `84030bbe3d74d99bad477a3d2e37a973ccd8865c` / `0.1.dev1+g84030bbe3.empty`, release `98dff2a81d747d1dba01a47f939f48c3526d4206` / `0.29.0+empty`. Each lane totals **563 passed, 35 skipped, 1 xfailed** across its selected pytest invocations. Skips/xfails are not passes. The resulting rebased integration commit is not printed and is not inferred. Actual release CPU and failed/cancelled image variants remain gaps. - Current-head CI update (2026-09-20): [main CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654/job/106062951955) passed **5157 tests**, with **67 skipped**; actual installed vLLM was `0.1.dev1+g84030bbe3.empty`. Ascend checkout was the current PR head; the log does not print the full resulting integration head/base. Pre-commit and mypy passed. NPU/E2E results are recorded above; actual release CPU remains unverified. - Image build is partially blocked by infrastructure: A5 amd64 [Ubuntu](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362005) and [openEuler](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362068) failed before reading the Dockerfile because BuildKit could not create a snapshot temporary directory (`no space left on device`). No release compatibility code change is justified by this failure; cancelled variants remain unverified. The successful [310P openEuler arm64 build](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062361987) explicitly checked out release `98dff2a81d747d1dba01a47f939f48c3526d4206` and installed `0.29.0+empty`; this is build evidence, not runtime UT coverage. - Current head: `a3b76b073a201851454e873aa89fc2254992fc06`, rebased onto the baseline above. Previous-head CPU [job 106060931458](https://github.com/vllm-project/vllm-ascend/actions/runs/35503843936/job/106060931458) stopped during Ascend integration rebase after #16803 merged, before any UT ran. Its vLLM checkout was the fixed main and installation reported `0.1.dev1+g84030bbe3.empty`. The two import conflicts are resolved; fresh-head CI was retriggered by the push. - Remote official v0.29.0 tag resolved to the exact SHA above; both source pins inspected. - All 95 changed Python files pass Ruff lint, Ruff format and AST parsing; `git diff --check` passes. Dockerfile tag/marker consistency checked across all eight variants. - Full pre-commit invocation: Ruff, codespell, typos, clang-format, markdownlint, actionlint, package-init, forbidden-import and boolean-context checks passed. Bash-dependent hooks cannot run on this Windows host; Python launcher hooks exit 9009. Full CI lint remains pending. - Actual main/release CPU/NPU and image builds are not claimed locally: the required Linux/NPU/container environment is unavailable. New PR E2E and image-build CI are requested. Existing CPU workflow runs fixed main only, so actual release CPU remains a validation gap. - #16393's historical green CI is not a substitute for this new head. No new test functions, workflow changes, golden/threshold changes or additional skips are introduced beyond restoring #16393. Inherited 0.28-only tests/skips are not counted as passes. - vLLM main: vllm-project/vllm@84030bb Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). Scope is the MRV2 composition only; MRV1 PP+SP pieces from the original branch are intentionally excluded. - Kimi K3 targets (KimiLinear / KimiK3 Causal + ConditionalGeneration) join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding: first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; MRV2 keeps local-sized receive/graph views over persistent buffer storage, and both MRV2 dummy paths (memory profiling and graph capture) slice the PP receive views. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU spec-token positions resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Adapted to main since vllm-project#16511: union of the dspark PP architecture table with GLM-5.2, execute_model nests the PP+SP buffer handling under pcp_dispatch_context with the 0.29.0+ dummy-state signature, and the ACL graph capture keeps main's pcp_manager flow inside the PP+SP capture-buffer context. CPU regression sandboxes gained the new execute_model free names (has_kv_transfer_group, pcp_dispatch_context, use_updatable_graph) and no longer extract the refactored run_fullgraph.

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747): common runner and graph files stay untouched, and every K3 PP+SP hook lives in a dedicated patch module gated by enable_kimi_k3_sp, so other models keep the upstream paths. - Kimi K3 targets (KimiLinear / KimiK3 Causal + ConditionalGeneration) join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather. - patch_kimi_k3_pp_sp.py narrows the persistent PP receive backing, graph capture/replay views and dummy-batch attention preparation to the local shard; keeps CPU attention lengths synchronized on every PP stage after speculative rejection and non-final prefill chunks. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. Adapted to main since vllm-project#16511: union of the dspark PP architecture table with GLM-5.2, the seq-lens hook follows main's _update_seq_lens_cpu layout, and the ACL graph shim sits between ModelWithContext and the model so the common wrapper keeps its upstream signature.

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). K3 with expert parallelism and TP > 1 keeps sequence shards across PP stages (e.g. 8 of 64 rows at TP8), while upstream's PP receive backing, graph capture views and dummy-batch attention preparation slice by the full token count, so the runner and graph paths narrow those views to the local shard over persistent storage. - Kimi K3 targets (KimiLinear / KimiK3 Causal + ConditionalGeneration) join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; MRV2 keeps local-sized receive/graph views over persistent buffer storage, and both MRV2 dummy paths (memory profiling and graph capture) slice the PP receive views. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU attention lengths resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Adapted to main since vllm-project#16511: union of the dspark PP architecture table with GLM-5.2, execute_model nests the PP+SP buffer handling under pcp_dispatch_context with the 0.29.0+ dummy-state signature, and the ACL graph capture keeps main's pcp_manager flow inside the PP+SP capture-buffer context.

…oject#16393) ### What this PR does / why we need it? Upgrade the supported vLLM release dependency from **v0.28.0 to the official v0.29.0**, alongside the fixed vLLM main commit inherited from vllm-project#16216 (already merged). | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve vllm-project#15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [vllm-project#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted vllm-project#16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including vllm-project#15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> #### Per-file adaptation ledger Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend vllm-project#15747](https://github.com/vllm-project/vllm-ascend/pull/15747/files#diff-a170a42fe1e9275c999642b05c4a437d0c9104dd8ea19151796142909b00ab2a) (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | vllm-project#15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend vllm-project#16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | #52839; both lanes use the common PCP import. Rebase keeps current upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | #51718/#53896/#56078 plus #50323; remove v0.28.0 layouts/copy APIs and import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend vllm-project#16409's host-parameter-update revert and removes only the obsolete v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, #51031; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | #49811; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | #### Exact upstream evidence | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | #### Additional inherited contracts | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend vllm-project#16791](https://github.com/vllm-project/vllm-ascend/pull/16791/files) | both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend vllm-project#15098](https://github.com/vllm-project/vllm-ascend/pull/15098/files) | both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | #### Rebase and retired-fallback evidence | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [vllm-project#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [vllm-project#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> ### Does this PR introduce _any_ user-facing change? Yes. The supported release changes from vLLM v0.28.0 to v0.29.0, and Ascend PCP+DP is enabled on v0.29.0. The fixed main commit remains supported. The vllm-ascend package version is unchanged. Obsolete v0.28-only skips are removed where the contracts are now supported. No skip is migrated to v0.29.0, and no golden data or precision threshold is changed. Existing unrelated exclusions remain exclusions, not passes. ### How was this patch tested? Current-head validation: [run 35420046256](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256) completed successfully on `fd876c5269fb6b39afa04e5a52433505e8786e22`: 39 successful jobs and 6 skipped jobs, no failures. - Pre-commit and mypy passed. Fixed-main [CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105836507531): **5078 passed, 67 skipped, 17 warnings**. Actual vLLM checkout `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, installed `0.1.dev1+g84030bbe3.empty`; Ascend checkout is the PR head. CPU `base_sha` is empty; the log says HEAD is up to date, so no unprinted integration SHA is inferred. - All **32 NPU jobs passed**, 16 per lane. Raw-log session totals per lane: **562 passed, 35 skipped, 1 xfailed** (execution counts, not deduplicated cases). Actual checkouts: main `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, release `98dff2a81d747d1dba01a47f939f48c3526d4206`; installed versions `0.1.dev1+g84030bbe3.empty` and `0.29.0+empty`. Each log confirms Ascend head `fd876c5269fb6b39afa04e5a52433505e8786e22`, integration base `e139b7d573d3769fd1407d5027d7d4831f5469f0`, and HEAD is up to date. - PCP+DP and PCP+PP+MTP suite: [main A3 four-card part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105845372541) and [release A3 four-card part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35420046256/job/105845372595) both passed. This does not establish a root cause for historical HCCL failures. - Skipped and expected-failure cases are not passes. **Current-head release CPU verification remains missing**: the existing CPU job runs fixed main only. Historical release CPU results do not fill this gap. No workflow changes or additional skips were introduced. Latest rebase: head `fd876c5269fb6b39afa04e5a52433505e8786e22` onto upstream main `e139b7d573d3769fd1407d5027d7d4831f5469f0`. The old-head green results below are historical; new-head CI has completed successfully; see the verified results below. The fixed vLLM main and release pins remain unchanged. PR-wide Ruff, formatting and Python AST checks passed for 95 Python files; no workflow changes. Rebase reconciliation: - Preserve upstream [vllm-project#16853](https://github.com/vllm-project/vllm-ascend/pull/16853/files), `8f2e3fed73328148ea603ccfe5764542fa7cbbd2`, without migrating its 0.28-only validator/dispatch gates to 0.29. Both supported pins already call the PCP manager for dispatch token counts. Restore the version-helper import needed by the inherited method. Its inherited 0.28-only skipped tests are not counted as passes. - Adapt the existing unsupported-feature UT to preserve the upstream list: both supported pins delegate PCP checks to the manager via vLLM [#53853](https://github.com/vllm-project/vllm/pull/53853/files), `e376d45e82cb7e220da430e3179e81eb0922cf56`; no obsolete 0.28 string filtering is reinstated. - Adapt the existing KVPP allocation-entry UT from [vllm-project#15514](https://github.com/vllm-project/vllm-ascend/pull/15514/files), `b64959a11ea69a51409f64e4e68da8e0924dec42`, to the common allocation entry already used by both supported pins after vLLM #51718. Preserve cache-view assertions and the upstream dtype changes. No test functions or skips added. - vllm-project#16544 remains reverted by vllm-project#16905; its V4.1 compatibility workaround remains removed. Historical validation before this rebase: - **Current head `8c5d785931c495701bf1da8b5bc80b321ebd3cd0`:** [run 35350119603](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603) completed successfully: **40 successful jobs, 6 skipped jobs, no failures**. Pre-commit and mypy passed. Local Ruff lint/format and AST checks passed for all 95 changed Python files. - **Actual main CPU:** [job 105617611031](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105617611031) reports **5074 passed, 37 skipped**, 17 warnings. Logs confirm the exact PR head, vLLM checkout `84030bbe3d74d99bad477a3d2e37a973ccd8865c`, and installed `0.1.dev1+g84030bbe3.empty`. CPU `base_sha` is empty; no integration merge is inferred. - **Actual dual-version NPU:** All 32 NPU job logs were audited, 16 per lane. Fixed main installed `0.1.dev1+g84030bbe3.empty`; official release checkout `98dff2a81d747d1dba01a47f939f48c3526d4206` installed `0.29.0+empty`. Each lane reports **562 passed, 35 skipped, 1 xfailed** across pytest session summaries (execution counts, not deduplicated unique tests). Logs confirm checkout of this PR head and integration base `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`; the rebase step reports HEAD is up to date. - **PCP combinations:** The context-parallel suite containing PCP+DP and PCP+PP+MTP passes on both [main A3 part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105658864889) and [release A3 part4](https://github.com/vllm-project/vllm-ascend/actions/runs/35350119603/job/105658864659). - **Release CPU gap:** Current-head actual release CPU remains missing because the workflow has only a main CPU entry. Version-parameterized UTs and historical release CPU runs do not replace actual release installation. Historical run 35213834751 at head `2560da9c7c2628f1b92b9305a1e80c5856bce74f` passed 4923 tests with 37 skips per lane; its temporary CPU workflow was reverted. - **Backup:** vllm-project#16898 remains unchanged with no test CI triggered. Local branch `codex/backup-16393-before-16905` preserves previous green `7ad4c82ee`; its results are not used as current-head validation. Skipped/xfail cases and jobs are not counted as passes. Earlier intermittent HCCL errors have no proven root cause; no speculative fix is included. This run passes existing CI, but complete dual-version CPU validation remains outstanding. - vLLM main: vllm-project/vllm@84030bb --------- Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

…vllm-project#17004) ### What this PR does / why we need it? Reintroduce the vLLM v0.29.0 release upgrade from vllm-project#16393, reverted by vllm-project#16949, with container defaults aligned to the supported release. Based on upstream main `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; preserves later merged changes. The image/source mismatch is confirmed in [nightly job 106004497020](https://github.com/vllm-project/vllm-ascend/actions/runs/35482881089/job/106004497020): the image installed `vllm 0.28.0+empty` (tag `v0.28.0`) while upgraded Ascend failed importing `_get_packed_kv_cache_groups`. All eight root Dockerfiles still defaulted to v0.28.0. The reusable image workflow passes no `VLLM_TAG` override, so those defaults govern release builds. Changing only the release marker does not update those images. - Fixed vLLM main remains `84030bbe3d74d99bad477a3d2e37a973ccd8865c`. - Release changes from v0.28.0 (`2cf0a6915ce544dc493a0990f2ea38d81601128a`) to official v0.29.0 (`98dff2a81d747d1dba01a47f939f48c3526d4206`, rechecked against the remote tag). - Restore vllm-project#16393's source compatibility and existing UT changes by reversing vllm-project#16949, then reconcile current upstream changes. No main interface scan is rerun for this release-only upgrade. - Update `VLLM_TAG` in all eight Dockerfiles: default/A3/A5/310P, Ubuntu/openEuler. Preserve existing exact-commit build overrides. No workflow changes. | Current-base adaptation | Exact cause and evidence | Lane | Validation | | --- | --- | --- | --- | | Eight Dockerfile `VLLM_TAG` defaults and release marker | vllm-project#16393 changed the release contract but omitted image defaults; the linked nightly log proves installation of 0.28.0 and missing `_get_packed_kv_cache_groups`. | Release images | All eight defaults match marker; image-build CI requested, pending | | `worker/v2/attn_utils.py::get_kv_cache_spec` and existing `_make_mla_layer` UT fixture | Ascend [vllm-project#16755](https://github.com/vllm-project/vllm-ascend/pull/16755/files), `df3e755e98fba8c6a18f200c645e0c2050469bc3`, added an old-release field selector. vLLM [#51718](https://github.com/vllm-project/vllm/pull/51718/files), `8bdc70ec7b379279ec0152343239c2d50aced687`, replaced `compress_ratio` with `tokens_per_state`; both exact supported pins use the latter. Use the common field, retaining metadata and cache-view assertions. | Both | Source inspection and static checks passed; actual CI pending | | `attention/attention_v1.py` import conflict | Preserve `attention_transfer_window` from Ascend [vllm-project#15636](https://github.com/vllm-project/vllm-ascend/pull/15636/files), `5c80630f28f8529aa82716e58b981a78819ec429`, and graph changes from vllm-project#16798; keep the common relocated PCP import from vllm-project#16393. Do not restore the old compute-start import or unused weak-reference import. | Both | Conflict resolved; static checks passed | | `worker/v2/aclgraph_utils.py` import conflict | Preserve `ContextSource`, `UpdatableGraph`, and `use_updatable_graph` from Ascend [vllm-project#16798](https://github.com/vllm-project/vllm-ascend/pull/16798/files), `34bb51f93724c565362f5108f5226303e1b56cad`, while removing the obsolete 0.28 version selector restored by the revert. | Both | Conflict resolved; static checks passed | | Existing 310P and Mamba model-runner UT imports | Preserve hardware-profile imports and mocks from Ascend [vllm-project#16803](https://github.com/vllm-project/vllm-ascend/pull/16803/files), `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; remove only the unused old-release selector import. Hardware capability routing remains unchanged. | Both | Real rebase conflicts resolved; AST/Ruff/format passed; fresh CI pending | Newly merged changes were reviewed for version-contract impact: vllm-project#16775/vllm-project#16834 PCP metadata and routing, vllm-project#16923 A3 SFA, vllm-project#16426/vllm-project#16924/vllm-project#16955 custom ops, vllm-project#16081 MTP/SP, vllm-project#16669 xlite, vllm-project#15636/vllm-project#16747 transfer, vllm-project#16913 A5 pages, vllm-project#16798 graph updates, vllm-project#16755 GLM PD, vllm-project#16952 C8 config, vllm-project#16673 operator removal and vllm-project#16320 Kimi-K3 KV pool. Preserve these changes; no additional source-proven version branch was identified beyond the entries above. In particular, vllm-project#16747's `UniformTypeKVCacheSpecs.kv_cache_specs` and per-layer `block_size` exist in both exact pins; vllm-project#16320 adds Ascend connector hooks. CI remains necessary to validate runtime interactions. CI/documentation-only PRs are retained unchanged. <details> <summary>Inherited per-file compatibility evidence from vllm-project#16393</summary> The following source-contract ledger is inherited from vllm-project#16393. Any historical verification wording refers only to that earlier PR; it does not certify this new head. New-head validation is listed below. | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve vllm-project#15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [vllm-project#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted vllm-project#16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including vllm-project#15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> #### Per-file adaptation ledger Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend vllm-project#15747](https://github.com/vllm-project/vllm-ascend/pull/15747/files#diff-a170a42fe1e9275c999642b05c4a437d0c9104dd8ea19151796142909b00ab2a) (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | vllm-project#15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend vllm-project#16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | #52839; both lanes use the common PCP import. Rebase keeps current upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | #51718/#53896/#56078 plus #50323; remove v0.28.0 layouts/copy APIs and import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend vllm-project#16409's host-parameter-update revert and removes only the obsolete v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, #51031; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | #49811; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | #### Exact upstream evidence | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | #### Additional inherited contracts | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend vllm-project#16791](https://github.com/vllm-project/vllm-ascend/pull/16791/files) | both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend vllm-project#15098](https://github.com/vllm-project/vllm-ascend/pull/15098/files) | both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | #### Rebase and retired-fallback evidence | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [vllm-project#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [vllm-project#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> </details> ### Does this PR introduce _any_ user-facing change? Yes. The supported vLLM release and default container builds move to v0.29.0, while the fixed main remains supported. The vllm-ascend package version does not change. Existing 0.29-only PCP+DP compatibility is restored. ### How was this patch tested? - Final E2E result for head `a3b76b073a201851454e873aa89fc2254992fc06`: [run 35504646654](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654) succeeded (39 successful jobs, 6 skipped). Raw logs from all 32 selected NPU jobs confirm the requested Ascend head, integration base `2de71b594319bde52c8bded69eba154f50be8e75`, and the actual vLLM pins/installations: fixed main `84030bbe3d74d99bad477a3d2e37a973ccd8865c` / `0.1.dev1+g84030bbe3.empty`, release `98dff2a81d747d1dba01a47f939f48c3526d4206` / `0.29.0+empty`. Each lane totals **563 passed, 35 skipped, 1 xfailed** across its selected pytest invocations. Skips/xfails are not passes. The resulting rebased integration commit is not printed and is not inferred. Actual release CPU and failed/cancelled image variants remain gaps. - Current-head CI update (2026-09-20): [main CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654/job/106062951955) passed **5157 tests**, with **67 skipped**; actual installed vLLM was `0.1.dev1+g84030bbe3.empty`. Ascend checkout was the current PR head; the log does not print the full resulting integration head/base. Pre-commit and mypy passed. NPU/E2E results are recorded above; actual release CPU remains unverified. - Image build is partially blocked by infrastructure: A5 amd64 [Ubuntu](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362005) and [openEuler](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362068) failed before reading the Dockerfile because BuildKit could not create a snapshot temporary directory (`no space left on device`). No release compatibility code change is justified by this failure; cancelled variants remain unverified. The successful [310P openEuler arm64 build](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062361987) explicitly checked out release `98dff2a81d747d1dba01a47f939f48c3526d4206` and installed `0.29.0+empty`; this is build evidence, not runtime UT coverage. - Current head: `a3b76b073a201851454e873aa89fc2254992fc06`, rebased onto the baseline above. Previous-head CPU [job 106060931458](https://github.com/vllm-project/vllm-ascend/actions/runs/35503843936/job/106060931458) stopped during Ascend integration rebase after vllm-project#16803 merged, before any UT ran. Its vLLM checkout was the fixed main and installation reported `0.1.dev1+g84030bbe3.empty`. The two import conflicts are resolved; fresh-head CI was retriggered by the push. - Remote official v0.29.0 tag resolved to the exact SHA above; both source pins inspected. - All 95 changed Python files pass Ruff lint, Ruff format and AST parsing; `git diff --check` passes. Dockerfile tag/marker consistency checked across all eight variants. - Full pre-commit invocation: Ruff, codespell, typos, clang-format, markdownlint, actionlint, package-init, forbidden-import and boolean-context checks passed. Bash-dependent hooks cannot run on this Windows host; Python launcher hooks exit 9009. Full CI lint remains pending. - Actual main/release CPU/NPU and image builds are not claimed locally: the required Linux/NPU/container environment is unavailable. New PR E2E and image-build CI are requested. Existing CPU workflow runs fixed main only, so actual release CPU remains a validation gap. - vllm-project#16393's historical green CI is not a substitute for this new head. No new test functions, workflow changes, golden/threshold changes or additional skips are introduced beyond restoring vllm-project#16393. Inherited 0.28-only tests/skips are not counted as passes. - vLLM main: vllm-project/vllm@84030bb Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). K3 with expert parallelism and TP > 1 keeps sequence shards across PP stages (e.g. 8 of 64 rows at TP8), while upstream's PP receive backing, graph capture views and dummy-batch attention preparation slice by the full token count, so the runner and graph paths narrow those views to the local shard over persistent storage. - Kimi K3 targets (KimiLinear / KimiK3 Causal + ConditionalGeneration) join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py, AttnRes path used by K3 checkpoints): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; dense MLP layers gather and reduce-scatter around their TP-sharded weights while routed MoE keeps its SP path. Configurations without attention residuals keep the upstream forward unchanged. - MRV2 keeps local-sized receive/graph views over persistent buffer storage, and both MRV2 dummy paths (memory profiling and graph capture) slice the PP receive views. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU attention lengths resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Adapted to main since vllm-project#16511: union of the dspark PP architecture table with GLM-5.2, execute_model nests the PP+SP buffer handling under pcp_dispatch_context with the 0.29.0+ dummy-state signature, and the ACL graph capture keeps main's pcp_manager flow inside the PP+SP capture-buffer context.

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). K3 with expert parallelism and TP > 1 keeps sequence shards across PP stages (e.g. 8 of 64 rows at TP8), while upstream's V2 receive path slices by the full token count. Mirror MRV1's proven sync_and_slice_intermediate_tensors SP branch on the V2 runner: the persistent receive backing is narrowed to a local view around the upstream copy, and restored after each invocation. - Kimi K3 targets (KimiLinear / KimiK3 Causal + ConditionalGeneration) join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py, AttnRes path used by K3 checkpoints): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; dense MLP layers gather and reduce-scatter around their TP-sharded weights while routed MoE keeps its SP path. Configurations without attention residuals keep the upstream forward unchanged. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU attention lengths resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Graph capture is untouched: the sharding lives entirely in the runner's tensor handoff path, like MRV1. K3 SP+PP under FULL_DECODE_ONLY graphs is not covered by this port and needs separate validation.

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). Models in _SP_ACROSS_PP_ARCHITECTURES (the K3 family today) keep sequence shards across PP stages with expert parallelism and TP > 1 (e.g. 8 of 64 rows at TP8), while the V2 receive path slices by the full token count. Mirror MRV1's proven sync_and_slice_intermediate_ tensors SP branch on the V2 runner via the generic enable_sp_across_pp gate: the persistent receive backing is narrowed to a local view around the upstream copy, and restored after each invocation. - Kimi K3 targets join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py, AttnRes path used by K3 checkpoints): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; dense MLP layers gather and reduce-scatter around their TP-sharded weights while routed MoE keeps its SP path. Configurations without attention residuals keep the upstream forward unchanged. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU attention lengths resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Graph capture is untouched: the sharding lives entirely in the runner's tensor handoff path, like MRV1. SP+PP under FULL_DECODE_ONLY graphs is not covered by this port and needs separate validation.

…vllm-project#17004) Reintroduce the vLLM v0.29.0 release upgrade from vllm-project#16393, reverted by on upstream main `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; preserves later merged changes. The image/source mismatch is confirmed in [nightly job 106004497020](https://github.com/vllm-project/vllm-ascend/actions/runs/35482881089/job/106004497020): the image installed `vllm 0.28.0+empty` (tag `v0.28.0`) while upgraded Ascend failed importing `_get_packed_kv_cache_groups`. All eight root Dockerfiles still defaulted to v0.28.0. The reusable image workflow passes no `VLLM_TAG` override, so those defaults govern release builds. Changing only the release marker does not update those images. - Fixed vLLM main remains `84030bbe3d74d99bad477a3d2e37a973ccd8865c`. - Release changes from v0.28.0 (`2cf0a6915ce544dc493a0990f2ea38d81601128a`) to official v0.29.0 (`98dff2a81d747d1dba01a47f939f48c3526d4206`, rechecked against the remote tag). - Restore vllm-project#16393's source compatibility and existing UT changes by reversing vllm-project#16949, then reconcile current upstream changes. No main interface scan is rerun for this release-only upgrade. - Update `VLLM_TAG` in all eight Dockerfiles: default/A3/A5/310P, Ubuntu/openEuler. Preserve existing exact-commit build overrides. No workflow changes. | Current-base adaptation | Exact cause and evidence | Lane | Validation | | --- | --- | --- | --- | | Eight Dockerfile `VLLM_TAG` defaults and release marker | vllm-project#16393 changed the release contract but omitted image defaults; the linked nightly log proves installation of 0.28.0 and missing `_get_packed_kv_cache_groups`. | Release images | All eight defaults match marker; image-build CI requested, pending | | `worker/v2/attn_utils.py::get_kv_cache_spec` and existing `_make_mla_layer` UT fixture | Ascend [vllm-project#16755](https://github.com/vllm-project/vllm-ascend/pull/16755/files), `df3e755e98fba8c6a18f200c645e0c2050469bc3`, added an old-release field selector. vLLM [#51718](https://github.com/vllm-project/vllm/pull/51718/files), `8bdc70ec7b379279ec0152343239c2d50aced687`, replaced `compress_ratio` with `tokens_per_state`; both exact supported pins use the latter. Use the common field, retaining metadata and cache-view assertions. | Both | Source inspection and static checks passed; actual CI pending | | `attention/attention_v1.py` import conflict | Preserve `attention_transfer_window` from Ascend [vllm-project#15636](https://github.com/vllm-project/vllm-ascend/pull/15636/files), `5c80630f28f8529aa82716e58b981a78819ec429`, and graph changes from the old compute-start import or unused weak-reference import. | Both | Conflict resolved; static checks passed | | `worker/v2/aclgraph_utils.py` import conflict | Preserve `ContextSource`, `UpdatableGraph`, and `use_updatable_graph` from Ascend [vllm-project#16798](https://github.com/vllm-project/vllm-ascend/pull/16798/files), `34bb51f93724c565362f5108f5226303e1b56cad`, while removing the obsolete 0.28 version selector restored by the revert. | Both | Conflict resolved; static checks passed | | Existing 310P and Mamba model-runner UT imports | Preserve hardware-profile imports and mocks from Ascend [vllm-project#16803](https://github.com/vllm-project/vllm-ascend/pull/16803/files), `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; remove only the unused old-release selector import. Hardware capability routing remains unchanged. | Both | Real rebase conflicts resolved; AST/Ruff/format passed; fresh CI pending | Newly merged changes were reviewed for version-contract impact: GLM PD, vllm-project#16952 C8 config, vllm-project#16673 operator removal and vllm-project#16320 Kimi-K3 KV pool. Preserve these changes; no additional source-proven version branch was identified beyond the entries above. In particular, vllm-project#16747's `UniformTypeKVCacheSpecs.kv_cache_specs` and per-layer `block_size` exist in both exact pins; vllm-project#16320 adds Ascend connector hooks. CI remains necessary to validate runtime interactions. CI/documentation-only PRs are retained unchanged. <details> <summary>Inherited per-file compatibility evidence from vllm-project#16393</summary> The following source-contract ledger is inherited from vllm-project#16393. Any historical verification wording refers only to that earlier PR; it does not certify this new head. New-head validation is listed below. | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve vllm-project#15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [vllm-project#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted vllm-project#16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including vllm-project#15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | vllm-project#15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend vllm-project#16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [vllm-project#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [vllm-project#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> </details> Yes. The supported vLLM release and default container builds move to v0.29.0, while the fixed main remains supported. The vllm-ascend package version does not change. Existing 0.29-only PCP+DP compatibility is restored. - Final E2E result for head `a3b76b073a201851454e873aa89fc2254992fc06`: [run 35504646654](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654) succeeded (39 successful jobs, 6 skipped). Raw logs from all 32 selected NPU jobs confirm the requested Ascend head, integration base `2de71b594319bde52c8bded69eba154f50be8e75`, and the actual vLLM pins/installations: fixed main `84030bbe3d74d99bad477a3d2e37a973ccd8865c` / `0.1.dev1+g84030bbe3.empty`, release `98dff2a81d747d1dba01a47f939f48c3526d4206` / `0.29.0+empty`. Each lane totals **563 passed, 35 skipped, 1 xfailed** across its selected pytest invocations. Skips/xfails are not passes. The resulting rebased integration commit is not printed and is not inferred. Actual release CPU and failed/cancelled image variants remain gaps. - Current-head CI update (2026-09-20): [main CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654/job/106062951955) passed **5157 tests**, with **67 skipped**; actual installed vLLM was `0.1.dev1+g84030bbe3.empty`. Ascend checkout was the current PR head; the log does not print the full resulting integration head/base. Pre-commit and mypy passed. NPU/E2E results are recorded above; actual release CPU remains unverified. - Image build is partially blocked by infrastructure: A5 amd64 [Ubuntu](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362005) and [openEuler](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362068) failed before reading the Dockerfile because BuildKit could not create a snapshot temporary directory (`no space left on device`). No release compatibility code change is justified by this failure; cancelled variants remain unverified. The successful [310P openEuler arm64 build](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062361987) explicitly checked out release `98dff2a81d747d1dba01a47f939f48c3526d4206` and installed `0.29.0+empty`; this is build evidence, not runtime UT coverage. - Current head: `a3b76b073a201851454e873aa89fc2254992fc06`, rebased onto the baseline above. Previous-head CPU [job 106060931458](https://github.com/vllm-project/vllm-ascend/actions/runs/35503843936/job/106060931458) stopped during Ascend integration rebase after vllm-project#16803 merged, before any UT ran. Its vLLM checkout was the fixed main and installation reported `0.1.dev1+g84030bbe3.empty`. The two import conflicts are resolved; fresh-head CI was retriggered by the push. - Remote official v0.29.0 tag resolved to the exact SHA above; both source pins inspected. - All 95 changed Python files pass Ruff lint, Ruff format and AST parsing; `git diff --check` passes. Dockerfile tag/marker consistency checked across all eight variants. - Full pre-commit invocation: Ruff, codespell, typos, clang-format, markdownlint, actionlint, package-init, forbidden-import and boolean-context checks passed. Bash-dependent hooks cannot run on this Windows host; Python launcher hooks exit 9009. Full CI lint remains pending. - Actual main/release CPU/NPU and image builds are not claimed locally: the required Linux/NPU/container environment is unavailable. New PR E2E and image-build CI are requested. Existing CPU workflow runs fixed main only, so actual release CPU remains a validation gap. - vllm-project#16393's historical green CI is not a substitute for this new head. No new test functions, workflow changes, golden/threshold changes or additional skips are introduced beyond restoring vllm-project#16393. Inherited 0.28-only tests/skips are not counted as passes. - vLLM main: vllm-project/vllm@84030bb Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

…vllm-project#17004) ### What this PR does / why we need it? Reintroduce the vLLM v0.29.0 release upgrade from vllm-project#16393, reverted by vllm-project#16949, with container defaults aligned to the supported release. Based on upstream main `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; preserves later merged changes. The image/source mismatch is confirmed in [nightly job 106004497020](https://github.com/vllm-project/vllm-ascend/actions/runs/35482881089/job/106004497020): the image installed `vllm 0.28.0+empty` (tag `v0.28.0`) while upgraded Ascend failed importing `_get_packed_kv_cache_groups`. All eight root Dockerfiles still defaulted to v0.28.0. The reusable image workflow passes no `VLLM_TAG` override, so those defaults govern release builds. Changing only the release marker does not update those images. - Fixed vLLM main remains `84030bbe3d74d99bad477a3d2e37a973ccd8865c`. - Release changes from v0.28.0 (`2cf0a6915ce544dc493a0990f2ea38d81601128a`) to official v0.29.0 (`98dff2a81d747d1dba01a47f939f48c3526d4206`, rechecked against the remote tag). - Restore vllm-project#16393's source compatibility and existing UT changes by reversing vllm-project#16949, then reconcile current upstream changes. No main interface scan is rerun for this release-only upgrade. - Update `VLLM_TAG` in all eight Dockerfiles: default/A3/A5/310P, Ubuntu/openEuler. Preserve existing exact-commit build overrides. No workflow changes. | Current-base adaptation | Exact cause and evidence | Lane | Validation | | --- | --- | --- | --- | | Eight Dockerfile `VLLM_TAG` defaults and release marker | vllm-project#16393 changed the release contract but omitted image defaults; the linked nightly log proves installation of 0.28.0 and missing `_get_packed_kv_cache_groups`. | Release images | All eight defaults match marker; image-build CI requested, pending | | `worker/v2/attn_utils.py::get_kv_cache_spec` and existing `_make_mla_layer` UT fixture | Ascend [vllm-project#16755](https://github.com/vllm-project/vllm-ascend/pull/16755/files), `df3e755e98fba8c6a18f200c645e0c2050469bc3`, added an old-release field selector. vLLM [#51718](https://github.com/vllm-project/vllm/pull/51718/files), `8bdc70ec7b379279ec0152343239c2d50aced687`, replaced `compress_ratio` with `tokens_per_state`; both exact supported pins use the latter. Use the common field, retaining metadata and cache-view assertions. | Both | Source inspection and static checks passed; actual CI pending | | `attention/attention_v1.py` import conflict | Preserve `attention_transfer_window` from Ascend [vllm-project#15636](https://github.com/vllm-project/vllm-ascend/pull/15636/files), `5c80630f28f8529aa82716e58b981a78819ec429`, and graph changes from vllm-project#16798; keep the common relocated PCP import from vllm-project#16393. Do not restore the old compute-start import or unused weak-reference import. | Both | Conflict resolved; static checks passed | | `worker/v2/aclgraph_utils.py` import conflict | Preserve `ContextSource`, `UpdatableGraph`, and `use_updatable_graph` from Ascend [vllm-project#16798](https://github.com/vllm-project/vllm-ascend/pull/16798/files), `34bb51f93724c565362f5108f5226303e1b56cad`, while removing the obsolete 0.28 version selector restored by the revert. | Both | Conflict resolved; static checks passed | | Existing 310P and Mamba model-runner UT imports | Preserve hardware-profile imports and mocks from Ascend [vllm-project#16803](https://github.com/vllm-project/vllm-ascend/pull/16803/files), `d3f6b4b59ab55dc42af1f30f425c893b95af5ee2`; remove only the unused old-release selector import. Hardware capability routing remains unchanged. | Both | Real rebase conflicts resolved; AST/Ruff/format passed; fresh CI pending | Newly merged changes were reviewed for version-contract impact: vllm-project#16775/vllm-project#16834 PCP metadata and routing, vllm-project#16923 A3 SFA, vllm-project#16426/vllm-project#16924/vllm-project#16955 custom ops, vllm-project#16081 MTP/SP, vllm-project#16669 xlite, vllm-project#15636/vllm-project#16747 transfer, vllm-project#16913 A5 pages, vllm-project#16798 graph updates, vllm-project#16755 GLM PD, vllm-project#16952 C8 config, vllm-project#16673 operator removal and vllm-project#16320 Kimi-K3 KV pool. Preserve these changes; no additional source-proven version branch was identified beyond the entries above. In particular, vllm-project#16747's `UniformTypeKVCacheSpecs.kv_cache_specs` and per-layer `block_size` exist in both exact pins; vllm-project#16320 adds Ascend connector hooks. CI remains necessary to validate runtime interactions. CI/documentation-only PRs are retained unchanged. <details> <summary>Inherited per-file compatibility evidence from vllm-project#16393</summary> The following source-contract ledger is inherited from vllm-project#16393. Any historical verification wording refers only to that earlier PR; it does not certify this new head. New-head validation is listed below. | Reference | Commit | |---|---| | vllm-ascend rebase base | `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c` | | PR head | `8c5d785931c495701bf1da8b5bc80b321ebd3cd0` | | Fixed vLLM main | `84030bbe3d74d99bad477a3d2e37a973ccd8865c` | | Previous release: v0.28.0 | `2cf0a6915ce544dc493a0990f2ea38d81601128a` | | Target release: v0.29.0 | `98dff2a81d747d1dba01a47f939f48c3526d4206` | - Use common implementations where v0.29.0 and the fixed main share KV-cache layouts, Mamba copy/group APIs, PCP handling and speculative-decoding contracts. - Retain explicit `vllm_version_is("0.29.0")` branches for contracts that still differ, including RoPE, scheduler block snapshots, InputBatch, ReplaySSM, KV zeroing and DSpark PP handling. - Remove obsolete v0.28.0 compatibility and adapt existing test fixtures. Version detection uses package versions and the explicit `VLLM_VERSION` override, with local-version suffix handling and UT environment isolation retained; no hard-coded release-SHA inference remains. - Apply the Ascend PCP+DP validator patch only to v0.29.0, preserving other validations and EPLB platform binding. Rebuild dependent Pydantic schemas so nested configuration validation uses the patched validator. The global patch documentation records its rationale and removal criteria. - Preserve vllm-project#15747's Spec+PP protocol/partition handling after rebase; use the common exact-release selector and the real function-local DSpark sharing import. This release-only upgrade does not require a new main old-to-new interface scan. The latest rebase incorporates [vllm-project#16905](https://github.com/vllm-project/vllm-ascend/pull/16905/files), which reverted vllm-project#16544 at `9dc6704559ebe2809b5cb7b7fc44184bd1338b3c`. The now-unnecessary V4.1 drafter import gate and tuple annotation have been removed. Other release adaptations, including vllm-project#15747 Spec+PP handling, remain. Detailed contract evidence is retained below for review. <details> <summary>Per-file adaptations and exact upstream evidence</summary> #### Per-file adaptation ledger Evidence IDs refer to the exact source contract and upstream diff table below. Every row is syntax checked; branch-normalized AST comparison confirms unchanged main function bodies except the version identity helper and the explicitly retired propose argument/type annotations. Both supported versions completed CPU and NPU execution as recorded below. | Ascend file / symbols | Disposition and upstream evidence | Verification | |---|---|---| | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` / `_load_dspark_model_with_target_quant`; `tests/ut/patch/worker/test_patch_dspark_pp.py` | Preserve rebased [Ascend vllm-project#15747](https://github.com/vllm-project/vllm-ascend/pull/15747/files#diff-a170a42fe1e9275c999642b05c4a437d0c9104dd8ea19151796142909b00ab2a) (`82df9d871`), including manual PP partition masking. vLLM [#52809 diff](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`91a893de64722019ea2faf852e06cabe143b3490`) moved `_should_share` inside the loader on both supported pins, so retain the earlier release fix: patch `eagle_utils`, never read/patch a nonexistent `dspark_utils._should_share`. v0.29 keeps its global PP guard; main has native PP via [#50514](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) (`d87a440f88e28e5b37f9b1e22ce214d0426d5352`). | Existing UT now checks the real function-local import, absent module alias, and restoration on success/failure for both lanes; partition and PP assertions retained. Ruff/syntax pass; actual CPU/NPU pending. | | `vllm_ascend/worker/v2/pp_utils.py` / `use_legacy_spec_pp`; `tests/ut/worker/v2/test_pp_utils.py` | vllm-project#15747 added broad 0.28/0.29 routing and an obsolete 0.28 dev-build recognition path. For the two supported pins, #50514 exists only on fixed main. Route through `vllm_version_is("0.29.0")`; retain package/local-suffix and explicit environment-override semantics. No release-SHA inference or third release lane. | Existing UTs exercise the real uncached version helper with monkeypatch isolation, local suffix, fixed-main dev string, explicit override, and non-target versions. Isolated routing checks and Ruff/syntax pass; full CPU UT pending. | | `vllm_ascend/_310p/model_runner_310p.py` `_prepare_inputs`, `_allocate_kv_cache_tensors` | #51718, #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/spec_decode/llm_base_proposer_310.py` `set_inputs_first_pass` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/model_runner.py` `initialize_kv_cache`, `_allocate_kv_cache_tensors`, `_prepare_inputs_310p` | #51718/#54436; use shared standardized layouts and retain the v0.29.0 InputBatch gate. The rebase preserves vllm-ascend vllm-project#16043's MTP copy tracking while removing only legacy v0.28.0 allocation paths. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/_310p/worker/v2/rope.py` `get_310p_rope_state` | #56078; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/attention_v1.py` `module imports/dispatch` | #52839; both lanes use the common PCP import. Rebase keeps current upstream graph code and removes only the v0.28.0 import branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/context_parallel/sfa_cp.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/indexer.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/attention/mla_v1.py` `module imports/dispatch` | #52839; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/dyntra_lb_scheduler.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/recompute_scheduler.py` `module imports/dispatch`, `schedule` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/scheduler_profiling_chunk.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/core/kv_cache_interface.py` `get_kv_cache_compression_ratio`, `AscendMLAAttentionSpec`, `get_storage_block_size` | #51718, #53906; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/ascend_store/layerwise_cache_layout.py` `_merge_specs` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/preempt_offload/manager.py` `_derive_cpu_config` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/distributed/kv_transfer/kv_pool/kv_offload/native/npu.py` `create_worker` | #52615; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_mtp.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/deepseek_v4/indexer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/cache_config.py` `make_tensor` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/glm5next/kv_cache.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/kimi_k3_dspark.py` `load_weights` | #53106; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/models/layer/attention/layer.py` `get_kv_cache_spec` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_balance_schedule.py` `module imports/dispatch` | #51358, #54853; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_coordinator.py` `__init__` | #53614; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_kv_cache_utils.py` `module imports/dispatch`, `_get_max_layers_per_page_size`, `_ascend_max_memory_usage_bytes_from_groups`, `_ascend_get_kv_cache_config_from_groups` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/platform/patch_use_v2_model_runner.py` `module imports/dispatch`, `_patched_get_unsupported_features` | #53853, #53183; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py` `bind_kv_cache` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_bind_kv_cache.py::bind_kv_cache` | v0.29.0 has no ReplaySSM ring-tracker helper; fixed main requires it after KV binding. Evidence: [#52506, `adebc41b7e9f1085d3f73434e23beb76883b9eb4`, worker-utils diff](https://github.com/vllm-project/vllm/pull/52506/files#diff-645d58630d5acf3a0b07226bfef1e890a584c32502ab97c3d4642070f39a783c). Keep an explicit v0.29.0 exclusion and preserve the fixed-main call. | Failure reproduced on v0.29.0; Historical release/main NPU validation passed; current-head CI pending | | `vllm_ascend/patch/worker/patch_mamba_utils.py` `module imports/dispatch`, `_get_state_copy_funcs_for_layer` | #53896; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_attn_utils.py` `module imports/dispatch` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/patch/worker/patch_v2/patch_dspark.py` `_load_dspark_model_with_target_quant` | #50514, #52809; v0.29.0 keeps `get_pp_group` module-bound but imports `_should_share` locally from Eagle utilities; fixed main removes the PP guard. Keep the release `get_pp_group` patch, share through the common Eagle utility, and delete the obsolete v0.28.0 `dspark_utils._should_share` patch. | Exact release failure reproduced; Historical release/main CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/spec_decode/llm_base_proposer.py` `model_returns_tuple`, `__init__`, `set_inputs_first_pass` | #56078, #52861; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/utils.py` `get_kv_cache_tensor_layers`, `register_ascend_customop`, `vllm_version_is` | #52839, #51718, #52494; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/model_runner_v1.py` `_get_ascend_mamba_state_copy_funcs`, `_allocate_kv_cache_tensors`, `_prepare_inputs`, `_dummy_run`, `_reshape_kv_cache_tensors`, `get_kv_cache_spec`, NaN helpers | #51718/#53896/#56078 plus #50323; remove v0.28.0 layouts/copy APIs and import the NaN helpers directly because v0.29.0 and fixed main expose the same contract. XD-RoPE remains explicitly gated to v0.29.0. | Static checked; historical dual-version CPU/NPU validation passed; current-head CI pending | | `vllm_ascend/worker/v2/aclgraph_utils.py` `capture` | #53869 makes `pcp_manager` common to both lanes. The rebase preserves vllm-ascend vllm-project#16409's host-parameter-update revert and removes only the obsolete v0.28.0 capture branch. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/attn_utils.py` `get_kv_cache_spec`, `_allocate_kv_cache`, `_reshape_kv_cache_v2` | #51718; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/block_table.py` `__init__`, `init_block_table_layout_tensors`, `compute_slot_mappings` | #51718, #51031; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/model_runner.py` `module imports/dispatch`, `prepare_inputs`, `__init__`, `sample_tokens`, `prepare_dummy_attn`, `execute_model` | #50514, #54436, #52506, #55212, #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/pcp_manager.py` `partition_batch` | #53515; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/sample/gumbel.py` `module imports/dispatch` | #54282; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/__init__.py` `init_speculator` | #49811; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dflash/speculator.py` `module imports/dispatch`, `propose` | #53694, #52188; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/v2/spec_decode/dspark/speculator.py` `propose` | #53694; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | | `vllm_ascend/worker/worker.py` `_scale_kv_cache_memory_for_multi_group`, `wake_up` | #51718, #53508; common contracts merged, differing release contracts explicitly gated as detailed below. | Static checked; prior-head dual-version results below; rebased CI pending | #### Exact upstream evidence | Upstream change | Full commit SHA / direct diff | Actual supported contracts and branch decision | |---|---|---| | #51718 | `8bdc70ec7b379279ec0152343239c2d50aced687` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/51718/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Both use layers/layer_stride/block_stride/offset, tokens_per_state, CircularBufferSpec and standardized backing; retire shared_by/compress_ratio allocation branches. | | #52839 | `58e5ee0158b6a264c3506f00480e108a34b33ee3` [vllm/v1/attention/ops/pcp.py](https://github.com/vllm-project/vllm/pull/52839/files#diff-23d7e7ab01a7f44414796e1e1b08cfa5277a05f54143ee2dd3a1041ca47772a8) | Both import PCP operations from vllm.v1.attention.ops.pcp. | | #53896 | `e126687a9a828d513c01a07cd69f025f27d63280` [vllm/v1/worker/mamba_utils.py](https://github.com/vllm-project/vllm/pull/53896/files#diff-8ab86225fa08e9a6700851191abcb3f4f8b0bba40e97a82aad700d7a23a4d7ec) | Both use Mamba copy-function dictionaries and unwrap UniformTypeKVCacheSpecs. | | #53106 | `1fe3a1571ac67581478a11743e55a306de1d136f` [vllm/model_executor/models/utils.py](https://github.com/vllm-project/vllm/pull/53106/files#diff-b0ba1095e9881e5c87e33dfd20958d1e1ceafe8a4433aa692f468e61be130b21) | Both use WeightsMapper instead of AutoWeightsLoader skip_prefixes/skip_substrs. | | #53906 | `98ed0856f31fa3aaf5e27464e2b4ef5a8ee6b2f5` [vllm/v1/kv_cache_interface.py](https://github.com/vllm-project/vllm/pull/53906/files#diff-f76cdfbf02dacd9dffbfbd0d9ad68a7a6ac0d8aed70834f95a2ae8ccd2e333cb) | Only pinned main has the optional MLA storage_block_size dataclass field; release keeps the Ascend derived property, using tokens_per_state. | | #56078 | `719284fe158f1be8a9dd92953295fc9d49015730` [vllm/config/model.py](https://github.com/vllm-project/vllm/pull/56078/files#diff-998c640befaf137b9af825f29f4e6e47d273caab1fd04093c97df24b18f5c417) | Release retains uses_xdrope_dim and three M-RoPE dimensions; pinned main uses mrope_num_dims and unified RoPE. | | #52615 | `138d137b5b955a5ebee98dfc946ecdd65d7b87ce` [vllm/v1/kv_offload/cpu/spec.py](https://github.com/vllm-project/vllm/pull/52615/files#diff-7d073c26c9b22f74e9ff9e4c8da733d1807233edb83439fc64fbd48c43ae6045) | Release uses num_blocks/kv_bytes_per_block; main uses num_chunks/kv_bytes_per_chunk. | | #51358 | `6b110badbb22d3f66c7218b71138f13b7a6b3419` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/51358/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release now has boundary_state_offloads and KVConnectorBlockState; remove partial_tail_offloads plumbing. | | #54853 | `0b066293f3c738a0cbd3a087bf893f2f4dcd61f2` [vllm/v1/core/sched/output.py](https://github.com/vllm-project/vllm/pull/54853/files#diff-cafd89ce8a698a56acb24ada62831cbc7a980782f78a52d1742ba238031f296c) | Release constructor takes block_ids snapshots; main takes req_ids and resolve_block_ids. Keep exact release snapshot membership and main lazy-resolution membership. | | #53614 | `144e79c8106da23141ac010394b782f730cc7fe8` [vllm/v1/core/kv_cache_coordinator.py](https://github.com/vllm-project/vllm/pull/53614/files#diff-43875c71daa893ef7567e21633d9988c2baf95bef61e3a334a6d584d6444d725) | Only main configures drop_eagle_checkpoint_block for replay-aligned Mamba checkpoints. | | #50514 | `d87a440f88e28e5b37f9b1e22ce214d0426d5352` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/50514/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Release retains module-level PP/share symbols and Ascend PP workaround; main has the subsequent PP integration. | | #52809 | `91a893de64722019ea2faf852e06cabe143b3490` [vllm/v1/worker/gpu/spec_decode/dspark/utils.py](https://github.com/vllm-project/vllm/pull/52809/files#diff-2223c5e6b44b302db9bff4023777cc88b4427f4fb9d4a04476f6d6e4bd687393) | Between v0.28.0 and v0.29.0, `_should_share` moved from a DSpark module binding to a function-local import from Eagle utilities. The shared Eagle patch remains effective; the old DSpark-module read/write must be removed. | | #54436 | `6bafc049aae6c26e210162630914ee9177a4b586` [vllm/v1/worker/gpu/input_batch.py](https://github.com/vllm-project/vllm/pull/54436/files#diff-106f39c08266f186830bb8fcd7fb1df35c0aaa5fd0ac5c17aac64aeddee48902) | Release InputBatch requires max_seq_len_np; main removed it. | | #52506 | `adebc41b7e9f1085d3f73434e23beb76883b9eb4` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52506/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Only main accepts valid_dummy_state_slots/valid_state_slots capture arguments. | | #55212 | `83990f5bcc0b5eb08b2f1fd1b109fa8a31b122f3` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/55212/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Release prepares DCP local sequence lengths before partitioning; main initializes DCP metadata afterwards. | | #53515 | `b1fbbc2ade51e3826bc92e4733c9c692ee21d42d` [vllm/v1/worker/gpu/pcp_manager.py](https://github.com/vllm-project/vllm/pull/53515/files#diff-45dad5fe975c2ec8d02124fa65579cf11bfe9169a94217ceae1de0ea29518127) | Both accept padded_num_tokens for persistent PCP input buffers. | | #53869 | `b3af042abd8fe5a297ec3ec72db276fd661a67b3` [vllm/v1/worker/gpu/cudagraph_utils.py](https://github.com/vllm-project/vllm/pull/53869/files#diff-fc699ff1c69fd17adb60116b3a7a20aa88b5187205f589952d0f3792475ec212) | Both accept pcp_manager during graph capture. | | #51031 | `0ecc284790e5403f74b899524ef82ecb69f83cb3` [vllm/v1/worker/gpu/block_table.py](https://github.com/vllm-project/vllm/pull/51031/files#diff-21052649468f36e592c9ed378a9cd7b4615c558fb4d10f034f7b4268d3b6f9e1) | Both distinguish KV and kernel block sizes during DCP slot mapping. | | #54282 | `fe755c88995ad468882517b6c4bdd60138d46a3a` [vllm/v1/worker/gpu/sample/gumbel.py](https://github.com/vllm-project/vllm/pull/54282/files#diff-02f7c5a06a2d23d544a07d57af16c7ffb74886469c8844385f51a719e329e6b8) | Both gumbel sampling APIs include is_drafting. | | #52188 | `d1e3eee6fb8ed3623241ef5c8e3ac533f775bff9` [vllm/v1/worker/gpu/spec_decode/dflash/speculator.py](https://github.com/vllm-project/vllm/pull/52188/files#diff-0220b514682125dea26026db2d7caa1a0d9c772d84452e9e124434d1e11be5f0) | Both DFlash input kernels take cp_rank/CP_SIZE/CP_INTERLEAVE. | | #53694 | `5acc1c4e4b8730298cbff7a4a7c68c814dc24fd7` [vllm/v1/worker/gpu/spec_decode/autoregressive/speculator.py](https://github.com/vllm-project/vllm/pull/53694/files#diff-3e7e2ba21d64e309503b7c5fe537363a7669fa1baf84d3c2e4891d9aed64fbe6) | Both propose APIs take DPSyncState; remove the obsolete token-count argument and retain replicated-PCP synchronization. | | #49811 | `01af92e175407231b1433b0aef01a1b9c983d955` [vllm/v1/worker/gpu/spec_decode/extract_hidden_states.py](https://github.com/vllm-project/vllm/pull/49811/files#diff-b13061cd1be0bc80aaa9fce208fe9128aaaad25b3832c5f1693da1d4e8263f5e) | Both support extract_hidden_states on MRV2; remove old unsupported dispatch/skip. | | #53508 | `479eeb32d2b432dbb4e442fbd3f94ca2eca35d67` [vllm/v1/worker/gpu_model_runner.py](https://github.com/vllm-project/vllm/pull/53508/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29) | Both remove post_kv_cache_wake_up; retire release-only call. | | #52494 | `3ff4f02dfe69abc1a0375d1ea8d8d5cb25609fcc` [vllm/models/kimi_k3/amd/mla.py](https://github.com/vllm-project/vllm/pull/52494/files#diff-aad295e6de3607a46cea22f9f1cca768d4937dc5e1d3ca58105513358ed4e4c1) | Only main provides KimiK3MultiHeadLatentAttentionWrapper; retain release exclusion. | | #52861 | `b09bd69b5bf14911abef9a0e8e493b83c8a38fa6` [vllm/v1/spec_decode/llm_base_proposer.py](https://github.com/vllm-project/vllm/pull/52861/files#diff-56fcad87dae9192cb0ab1643473765096cc44e1e744dccf3b99c4f2d85360d28) | Both include DeepseekV32MTPModel in the two-hidden-state architecture set. | | #54713 | `b28c3e1568bfae930f61d4b24940e47528c85d4a` [vllm/v1/core/single_type_kv_cache_manager.py](https://github.com/vllm-project/vllm/pull/54713/files#diff-97c184a680b7a4bd7d58b11aa0073706533cc887d990eddb98469e7374025ab3) | Only main takes replay_boundaries in compressed-prefix hit lookup; preserve release calls without that keyword. | | #42785 | `442d36031ce710aa2a777c353becb660c57ab2bd` [vllm/v1/worker/encoder_cudagraph.py](https://github.com/vllm-project/vllm/pull/42785/files#diff-287435cc78f753ae42acda542bca01c094557cd17db6bc2991d968fd012c1f18) | Only main capture callers pass axis_keys; preserve the existing Ascend rejection of nonempty axes. | | #52358 | `8f816a3f665489d7f0d222115d4f72ebab01076b` [vllm/v1/worker/gpu/model_runner.py](https://github.com/vllm-project/vllm/pull/52358/files#diff-5823f988fc0264681a80db24ccaba4d364f14394815d83ec1d90944c09f571f0) | Both ExecuteModelState have dp_sync; only main has cudagraph_stats. | | #52789 | `9eb9d9d3953959695108600c8ed33d36bc6a1e5f` [vllm/v1/core/sched/scheduler.py](https://github.com/vllm-project/vllm/pull/52789/files#diff-9eeca590fd99f15621897e559dba39b3ec4e7c2c65ec3c3229711689e008b5f4) | Release already has mamba_has_prefill_checkpoint_blocks; later main also has fine-grained prefix-cache state. | | #51251 | `7bbbf7c8e5040f7ebd374e8cbc657e01af1136dd` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/51251/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | v0.29.0 and main expose ec_manager_config; retire the old release-only ScoreEncoder configuration skip. | | #53240 | `b2db227a7c4c5e55f85524a094b607e5d27408b4` [vllm/model_executor/layers/fused_moe/routed_experts_capturer.py](https://github.com/vllm-project/vllm/pull/53240/files#diff-0bbfc19dd02619bbcc948fa0281d4ab9a5b77b7d5062cde9911319c93089535c) | v0.29.0 supports MRV2 routed-expert capture and unwraps uniform cache groups; retire the old release-only replay skip. | | #53853 | `e376d45e82cb7e220da430e3179e81eb0922cf56` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53853/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both delegate PCP compatibility validation to the PCP manager. | | #53183 | `4aab2b0ebed20343efe543c633f71b3c1336d5b8` [vllm/config/vllm.py](https://github.com/vllm-project/vllm/pull/53183/files#diff-bee6813076031d3ca1edc903c1b02b81e4676519afc562ce3fefe37f20c7b650) | Both expose the V1 unsupported-feature helper used by the existing Ascend MRV1 feature filter. | #### Additional inherited contracts | vllm-ascend change | Why it is required | Upstream cause and direct link | Lane | Verification | |---|---|---|---|---| | `_310p/worker/v2/model_state.py`: remove tuple-return and RecoverSSM v0.28 fallbacks | v0.29.0 and fixed main both return `dict[MambaSpec, list[int]]` from `get_mamba_groups` and both initialize `recoverssm`; keeping the old fallback would preserve an unsupported third contract | [v0.29.0 mamba groups](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/mamba_utils.py#L691-L704), [fixed-main mamba groups](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/mamba_utils.py#L691-L704), [v0.29.0 RecoverSSM](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103), [fixed-main RecoverSSM](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_states/mamba_hybrid.py#L80-L103) | both, common implementation | Existing constructor UT now asserts the parent-created RecoverSSM value is retained; Ruff and compileall pass | | `worker/v2/model_runner.py`: always forward `kv_cache_allocation_context` | v0.29.0 and fixed main both accept this keyword; the new-base gate was specifically for v0.28.0 | [v0.29.0 signature](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/model_runner.py#L533-L538), [fixed-main signature](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/v1/worker/gpu/model_runner.py#L561-L566), [vllm-ascend vllm-project#16791](https://github.com/vllm-project/vllm-ascend/pull/16791/files) | both, common implementation | Existing UT continues to assert the exact context object reaches the parent; Ruff and compileall pass | | `_310p/worker/v2/model_runner.py`: select the release KV-zeroing contract with `vllm_version_is("0.29.0")` | v0.29.0 exposes `KVCacheGroupSpec.is_eagle_group` but lacks `SpeculativeConfig.use_eagle_block_drop`; fixed main added the method | vLLM [#53388 diff](https://github.com/vllm-project/vllm/pull/53388/files), commit [`481839ad9e5ebf87aecb54fa5c9d986bd5ea4b81`](vllm-project/vllm@481839a); [v0.29.0 group field](https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/kv_cache_interface.py#L1200), [fixed-main helper](https://github.com/vllm-project/vllm/blob/84030bbe3d74d99bad477a3d2e37a973ccd8865c/vllm/config/speculative.py#L1874-L1876) | release differs from main | Existing two-path KV-zeroing UT retained and renamed for v0.29.0; Ruff and compileall pass | | existing DFlash kernel UT: remove v0.28-only kwarg omission | the current Ascend kernel accepts the CP arguments and the only excluded lane was v0.28.0, which this PR replaces | [vllm-ascend vllm-project#15098](https://github.com/vllm-project/vllm-ascend/pull/15098/files) | both, common invocation | Existing NPU test remains enabled with all assertions; Current-head CI pending | | Ascend change | Why / upstream cause | Lane | Verification | |---|---|---|---| | `patch/platform/patch_parallel_config.py`, registration, and global patch documentation | Allow Ascend PCP+DP by removing the generic GPU restriction, following vLLM [#54523](https://github.com/vllm-project/vllm/pull/54523/files), commit `7c2f1ff4958eaf0818405e9192c71608fe4a16b1`. Preserve dynamic `parallel_config.current_platform` lookup and rebuild ParallelConfig → SpeculativeConfig → VllmConfig. | v0.29.0 only | Existing configuration/EPLB UTs and historical PCP+DP NPU execution passed; current-head CI pending. | | `ops/triton/v2/sample/categorical_sample.py`, existing Gumbel UT | Both supported contracts require `is_drafting`, from [#54282](https://github.com/vllm-project/vllm/pull/54282/files), `fe755c88995ad468882517b6c4bdd60138d46a3a`; preserve the inherited optimized kernel and use a common wrapper. | Both | Existing positive drafting assertion retained; current-head CI pending. | | `patch/platform/patch_kv_cache_coordinator.py`, existing prefix-cache UT | Both support `cache_hit_alignment_tokens`, introduced by [#53598](https://github.com/vllm-project/vllm/pull/53598/files), `2ba984a5d06db414f3b2474fe9338faf6cd80a1c`; retire the v0.28-only write-mask branch. | Both | Existing assertions retained; current-head CI pending. | #### Rebase and retired-fallback evidence | Change | Exact source evidence | Decision | |---|---|---| | `tests/e2e/conftest.py::PROMPT_CONFIGS[HunyuanOCR]` | vLLM [#53272](https://github.com/vllm-project/vllm/pull/53272/files#diff-0852f1e9753819abe4f85380abdc0660fd4a998f44bd7bb73d68f462bd776d95), `d53b1c2efc0ca7161c3844f51ca9acbcdb7129d5` removes native Hunyuan V1/VL before v0.29.0; both exact supported sources lack it. The old conditional came from vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-824bc3e5c28226a3c0f4581750fa93a7da7001aff9e0ff4aee4131e3edd7b144), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Remove the v0.28.0 selector and use the existing exclusion for both supported lanes. This does not migrate the protected SFA PCP skip or add a v0.29.0 accuracy skip. | | `vllm_ascend/worker/model_runner_v1.py` NaN helper imports and `tests/ut/worker/test_model_runner_v1_nan_detection.py` | vLLM [#50323](https://github.com/vllm-project/vllm/pull/50323/files#diff-80ee7e2a62f9dcfbb8a312dc4e3948557e97ef187290daebbcae1e28596bda29), `12292d94b25869be2af6b6d4f8eea6c2445e935f`; both exact sources export `nans_to_dict`, `gpu_sync_allowed`, and `raise_if_nan_logits`. The v0.28.0 fallback originated in vllm-ascend [vllm-project#14898](https://github.com/vllm-project/vllm-ascend/pull/14898/files#diff-c49594855b615477bbc34f06d2d423a7dd84c021a7925cd1f61fdb79cb814c08), `e5118d151314ae18e56c0b63aa8dd00d294adc22`. | Delete import-failure/`None` fallbacks and the existing UT's obsolete availability skip; assertions remain unchanged and execute on both lanes. | | `vllm_ascend/worker/v2/aclgraph_utils.py` | vllm-ascend [vllm-project#16409](https://github.com/vllm-project/vllm-ascend/pull/16409/files#diff-dc1c8a80dadf23627467bf04d6bc59fe13c6d075b7d2564d0acc07dc97d921f3), `799801feef347469d5e9b39374b9210e0d4f7431`. | Preserve the upstream revert while resolving the real rebase conflict; do not reintroduce the reverted graph-update behavior. | | `vllm_ascend/_310p/worker/v2/model_runner.py` | vllm-ascend [vllm-project#16043](https://github.com/vllm-project/vllm-ascend/pull/16043/files#diff-2f27611a7cfbbdb780b9e88157f05d14d6dfe65406c8d66a5ba1f6949124f5d2), `d4d2957e5208c2f464d4625c05920bd29ea233cb`. | Preserve the newly merged 310P MTP copy tracking and remove only the v0.28.0 descriptor/allocation branch. | </details> </details> ### Does this PR introduce _any_ user-facing change? Yes. The supported vLLM release and default container builds move to v0.29.0, while the fixed main remains supported. The vllm-ascend package version does not change. Existing 0.29-only PCP+DP compatibility is restored. ### How was this patch tested? - Final E2E result for head `a3b76b073a201851454e873aa89fc2254992fc06`: [run 35504646654](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654) succeeded (39 successful jobs, 6 skipped). Raw logs from all 32 selected NPU jobs confirm the requested Ascend head, integration base `2de71b594319bde52c8bded69eba154f50be8e75`, and the actual vLLM pins/installations: fixed main `84030bbe3d74d99bad477a3d2e37a973ccd8865c` / `0.1.dev1+g84030bbe3.empty`, release `98dff2a81d747d1dba01a47f939f48c3526d4206` / `0.29.0+empty`. Each lane totals **563 passed, 35 skipped, 1 xfailed** across its selected pytest invocations. Skips/xfails are not passes. The resulting rebased integration commit is not printed and is not inferred. Actual release CPU and failed/cancelled image variants remain gaps. - Current-head CI update (2026-09-20): [main CPU UT](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646654/job/106062951955) passed **5157 tests**, with **67 skipped**; actual installed vLLM was `0.1.dev1+g84030bbe3.empty`. Ascend checkout was the current PR head; the log does not print the full resulting integration head/base. Pre-commit and mypy passed. NPU/E2E results are recorded above; actual release CPU remains unverified. - Image build is partially blocked by infrastructure: A5 amd64 [Ubuntu](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362005) and [openEuler](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062362068) failed before reading the Dockerfile because BuildKit could not create a snapshot temporary directory (`no space left on device`). No release compatibility code change is justified by this failure; cancelled variants remain unverified. The successful [310P openEuler arm64 build](https://github.com/vllm-project/vllm-ascend/actions/runs/35504646536/job/106062361987) explicitly checked out release `98dff2a81d747d1dba01a47f939f48c3526d4206` and installed `0.29.0+empty`; this is build evidence, not runtime UT coverage. - Current head: `a3b76b073a201851454e873aa89fc2254992fc06`, rebased onto the baseline above. Previous-head CPU [job 106060931458](https://github.com/vllm-project/vllm-ascend/actions/runs/35503843936/job/106060931458) stopped during Ascend integration rebase after vllm-project#16803 merged, before any UT ran. Its vLLM checkout was the fixed main and installation reported `0.1.dev1+g84030bbe3.empty`. The two import conflicts are resolved; fresh-head CI was retriggered by the push. - Remote official v0.29.0 tag resolved to the exact SHA above; both source pins inspected. - All 95 changed Python files pass Ruff lint, Ruff format and AST parsing; `git diff --check` passes. Dockerfile tag/marker consistency checked across all eight variants. - Full pre-commit invocation: Ruff, codespell, typos, clang-format, markdownlint, actionlint, package-init, forbidden-import and boolean-context checks passed. Bash-dependent hooks cannot run on this Windows host; Python launcher hooks exit 9009. Full CI lint remains pending. - Actual main/release CPU/NPU and image builds are not claimed locally: the required Linux/NPU/container environment is unavailable. New PR E2E and image-build CI are requested. Existing CPU workflow runs fixed main only, so actual release CPU remains a validation gap. - vllm-project#16393's historical green CI is not a substitute for this new head. No new test functions, workflow changes, golden/threshold changes or additional skips are introduced beyond restoring vllm-project#16393. Inherited 0.28-only tests/skips are not counted as passes. - vLLM main: vllm-project/vllm@84030bb Signed-off-by: shenzhao <shenzhao9@huawei.com> Co-authored-by: shenzhao <shenzhao9@huawei.com>

Port the PP+SP+DSpark composition from the validated vllm-project#16511 onto current main, following the GLM-5.2 DSpark+PP layout (vllm-project#15747). Models in _SP_ACROSS_PP_ARCHITECTURES (the K3 family today) keep sequence shards across PP stages with expert parallelism and TP > 1 (e.g. 8 of 64 rows at TP8), while the V2 receive path slices by the full token count. Mirror MRV1's proven sync_and_slice_intermediate_ tensors SP branch on the V2 runner via the generic enable_sp_across_pp gate: the persistent receive backing is narrowed to a local view around the upstream copy, and restored after each invocation. - Kimi K3 targets join the dspark Spec+PP support table next to DSV4 and GLM-5.2, with aux hidden states and the upstream PP guard bypass. - PP+SP token sharding (kimi_k3.py, AttnRes path used by K3 checkpoints): first-stage-only embedding sharding, matching hidden/residual shards between stages, last-stage gather; dense MLP layers gather and reduce-scatter around their TP-sharded weights while routed MoE keeps its SP path. Configurations without attention residuals keep the upstream forward unchanged. - Local K3 MLA (K3DSparkModel) and legacy GQA (Qwen3DSparkModel) drafters validate as PP=1 on the last stage; cumulative aux-state relay keeps MLA drafts on raw prefix sums without the AttnRes bank. - CPU attention lengths resynchronize on every PP stage after speculative rejection and non-final prefill chunks. Graph capture is untouched: the sharding lives entirely in the runner's tensor handoff path, like MRV1. SP+PP under FULL_DECODE_ONLY graphs is not covered by this port and needs separate validation.

github-actions Bot added the module:tests label Sep 4, 2026

gemini-code-assist Bot reviewed Sep 4, 2026

View reviewed changes

Comment thread vllm_ascend/models/qwen3_dspark.py Outdated

LostFox11 force-pushed the codex/glm52-dspark-pp-mrv2 branch from 0728a57 to 447be90 Compare September 4, 2026 04:33

github-actions Bot added the merge-conflicts label Sep 8, 2026

LostFox11 requested a review from weijinqian0 as a code owner September 8, 2026 02:49

LostFox11 mentioned this pull request Sep 8, 2026

[BugFix][MRV2] Separate speculative PP sequence bounds from exact lengths #16007

Open

github-actions Bot added merge-conflicts and removed merge-conflicts labels Sep 8, 2026

github-actions Bot removed the merge-conflicts label Sep 9, 2026

LostFox11 force-pushed the codex/glm52-dspark-pp-mrv2 branch 3 times, most recently from b32eba4 to f5140f9 Compare September 9, 2026 17:46

LostFox11 requested review from ZT-AIA and zzzzwwjj as code owners September 10, 2026 01:54

LostFox11 force-pushed the codex/glm52-dspark-pp-mrv2 branch from 47323d6 to d017b7a Compare September 10, 2026 06:55

LostFox11 requested review from Yikun and wangxiyuan as code owners September 10, 2026 07:59

CXY-Katrina added the ready-precise run selected e2e test for pr label Sep 10, 2026

weiguihua2 reviewed Sep 10, 2026

View reviewed changes

Comment thread tests/e2e/pull_request/eight_card/model_runner_v2/test_glm5_2_spec_pp.py Outdated

LostFox11 commented Sep 10, 2026

View reviewed changes

Comment thread vllm_ascend/patch/worker/patch_v2/patch_spec_pp.py Outdated

LostFox11 commented Sep 10, 2026

View reviewed changes

Comment thread vllm_ascend/patch/worker/patch_deepseek_v2.py

LostFox11 mentioned this pull request Sep 10, 2026

[BugFix][SpecDecode] Align GLM DSpark auxiliary states with the upstream PP relay #16263

Open

LostFox11 force-pushed the codex/glm52-dspark-pp-mrv2 branch from dddaf1e to d3ae04a Compare September 17, 2026 08:51

nwpu-zxr approved these changes Sep 18, 2026

View reviewed changes

yiz-liu approved these changes Sep 18, 2026

View reviewed changes

yiz-liu merged commit 82df9d8 into vllm-project:main Sep 18, 2026
32 checks passed

zhao-stack mentioned this pull request Sep 18, 2026

[Misc][Main2Main] Upgrade vLLM release dependency to v0.29.0 #16393

Merged

zhao-stack mentioned this pull request Sep 20, 2026

[Misc][Main2Main] Upgrade vLLM to v0.29.0 with matching Docker images #17004

Merged

LostFox11 mentioned this pull request Sep 21, 2026

[Feature][Model] Support Kimi K3 PP+DSpark+SP on MRV2 #17157

Open

LQDLove mentioned this pull request Sep 22, 2026

[BugFix] Fix stale CPU UTs for vLLM 0.29.0 daily test #17234

Closed

Navigation Menu

[Feature][MRV2] Support GLM5.2 DSpark with PP - #15747

[Feature][MRV2] Support GLM5.2 DSpark with PP#15747
yiz-liu merged 1 commit into
vllm-project:mainfrom
LostFox11:codex/glm52-dspark-pp-mrv2

LostFox11 commented Sep 4, 2026 •

edited by github-actions Bot

Loading

Uh oh!

gemini-code-assist Bot commented Sep 4, 2026

Uh oh!

github-actions Bot commented Sep 4, 2026

💡 Consider Linking a Related Issue or RFC

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

Uh oh!

LostFox11 commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

LostFox11 commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

LostFox11 commented Sep 10, 2026 •

edited by vllm-ascend-ci

Loading

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

LostFox11 commented Sep 4, 2026 • edited by github-actions Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this PR does / why we need it?

Does this PR introduce any user-facing change?

How was this patch tested?

Hardware validation (169 server, 16× Ascend 910B)

CPU tests

Smoke test (169 native path)

Uh oh!

gemini-code-assist Bot commented Sep 4, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

github-actions Bot commented Sep 4, 2026

💡 Consider Linking a Related Issue or RFC

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

LostFox11 commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

LostFox11 commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026

Uh oh!

LostFox11 commented Sep 10, 2026 • edited by vllm-ascend-ci Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

LostFox11 commented Sep 4, 2026 •

edited by github-actions Bot

Loading

LostFox11 commented Sep 10, 2026 •

edited by vllm-ascend-ci

Loading